linux-kernel - Re: [PATCH 08/13] fs: limit filesystem stacking depth

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <87mx17du3w.fsf@tucsk.pomaz.szeredi.hu>
Date:	Mon, 03 Sep 2012 17:05:55 +0200
From:	Miklos Szeredi <miklos@...redi.hu>
To:	sedat.dilek@...il.com
Cc:	viro@...iv.linux.org.uk, linux-fsdevel@...r.kernel.org,
	linux-kernel@...r.kernel.org, hch@...radead.org,
	torvalds@...ux-foundation.org, akpm@...ux-foundation.org,
	apw@...onical.com, nbd@...nwrt.org, neilb@...e.de,
	jordipujolp@...il.com, ezk@....cs.sunysb.edu, ricwheeler@...il.com,
	dhowells@...hat.com, hpj@...la.net, sedat.dilek@...glemail.com,
	penberg@...nel.org, goran.cetusic@...il.com, romain@...bokech.com
Subject: Re: [PATCH 08/13] fs: limit filesystem stacking depth

Sedat Dilek <sedat.dilek@...il.com> writes:

>>
>> Yes, setting it to 2 is defensive, it would need more stack depth
>> analysis to see what an acceptable number would be.
>>
>
> Can you describe such an analysis method (in case you need help for
> testing it)?


I attached a systemtap script (x86-64 specific) which prints stack use
for stacked overlayfs filesystems.  Interpret output like this:

7288    0  ovl_lookup (bash/5721)
^          ^
|          |
|          +- function name (process/pid)
|
+------------ current stack use

7080  208          ovl_permission (bash/5721)
      ^    ^
      |    |
      |    +- stacking depth indicated by indentation
      |
      +------ stack increase from previous stacking level


You can try it on various setups (overlayfs being used as the lower
and/or upper level) and executing various filesystem operations.

Looks like "copy up" is the most stack hungry operation, it may be worth
trying to reduce its stack usage.

Thanks,
Miklos
---

global rec_level, stacks

probe
	kernel.function("ovl_permission"),
	kernel.function("ovl_getattr"),
	kernel.function("ovl_dir_getattr"),
	kernel.function("ovl_setattr"),
	kernel.function("ovl_setxattr"),
	kernel.function("ovl_listxattr"),
	kernel.function("ovl_removexattr"),
	kernel.function("ovl_dentry_open"),
	kernel.function("ovl_lookup"),
	kernel.function("ovl_mkdir"),
	kernel.function("ovl_symlink"),
	kernel.function("ovl_unlink"),
	kernel.function("ovl_rmdir"),
	kernel.function("ovl_rename"),
	kernel.function("ovl_link"),
	kernel.function("ovl_create"),
	kernel.function("ovl_mknod"),
	kernel.function("ovl_follow_link"),
	kernel.function("ovl_put_link"),
	kernel.function("ovl_readlink"),
	kernel.function("ovl_dir_open"),
	kernel.function("ovl_readdir"),
	kernel.function("ovl_dir_llseek"),
	kernel.function("ovl_dir_fsync"),
	kernel.function("ovl_dir_release"),
	kernel.function("ovl_dentry_release"),
	kernel.function("ovl_put_super"),
	kernel.function("ovl_statfs")
{
	tid = tid();
	i = rec_level[tid]++;
	stack_rem = u_register("rbp") & 0x1fff;
	stacks[tid, i] = stack_rem;
	delta = i > 0 ? stacks[tid, i - 1] - stack_rem : 0;
	printf("%4i %4i %-*s %s (%s/%i)\n", stack_rem, delta, i * 8, "", probefunc(), execname(), tid);
}

probe
	kernel.function("ovl_permission").return,
	kernel.function("ovl_getattr").return,
	kernel.function("ovl_dir_getattr").return,
	kernel.function("ovl_setattr").return,
	kernel.function("ovl_setxattr").return,
	kernel.function("ovl_listxattr").return,
	kernel.function("ovl_removexattr").return,
	kernel.function("ovl_dentry_open").return,
	kernel.function("ovl_lookup").return,
	kernel.function("ovl_mkdir").return,
	kernel.function("ovl_symlink").return,
	kernel.function("ovl_unlink").return,
	kernel.function("ovl_rmdir").return,
	kernel.function("ovl_rename").return,
	kernel.function("ovl_link").return,
	kernel.function("ovl_create").return,
	kernel.function("ovl_mknod").return,
	kernel.function("ovl_follow_link").return,
	kernel.function("ovl_put_link").return,
	kernel.function("ovl_readlink").return,
	kernel.function("ovl_dir_open").return,
	kernel.function("ovl_readdir").return,
	kernel.function("ovl_dir_llseek").return,
	kernel.function("ovl_dir_fsync").return,
	kernel.function("ovl_dir_release").return,
	kernel.function("ovl_dentry_release").return,
	kernel.function("ovl_put_super").return,
	kernel.function("ovl_statfs").return
{
	rec_level[tid()]--;
}


--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/